Investigating Tradeoffs between Practice and Observation in Automated Instruction

This study examined the performance to observation ratio in training a complex computer-based flight simulator task. Subjects were divided into five different groups. One group performed 100% of the time during training, the second group performed 75% of the time and observed 25% of the time, the third group performed 50% of the time and observed 50% of the time, the fourth group performed 25% of the time and observed 75%, and the fifth group observed 100% of the time. Based on previous observational learning literature, we predicted a linear relationship between performance and percent of time performing. Results however, suggest a non-linear relationship between performance and percent of time performing. Performance was slightly better when a combination of performance and observation was used rather than performance alone. These results indicate that observational learning plays an important, yet often neglected role in learning complex computer-based tasks and suggests that small group oriented computer-based training systems may be more instructionally and cost effective than individually oriented computer-based training systems.